Skip to main content
Incident intelligence/SS-IR-068CASE FILE OPEN
Symbolic editorial illustration for SS-IR-068SERVANTSTACK // INCIDENT INTELLIGENCEFORENSIC IMAGE // VERIFIED FRAME
SS-IR-068 // INCIDENT REPORTReported

Waymo

Robotaxis Drove Past Stopped School Buses 20 Times, Triggering a 3,067-Vehicle Recall

EXECUTIVE BRIEF

During the 2025-2026 school year, the Austin Independent School District documented roughly 20 separate instances of driverless Waymo robotaxis illegally passing stopped school buses that had their red lights flashing and stop arms extended.

AUTONOMOUS & AVIATION
FAILURE CHAINTRACE COMPLETE
  1. 01TRIGGERDuring the 2025-2026 school year, the Austin Independent School District documented roughly 20 separate instances of…
  2. 02MACHINE ACTIONAutonomous actor
  3. 03MISSING GATEExecution gate and human override
  4. 04IMPACTPhysical safety
01 // INCIDENT SUMMARY

The short version

During the 2025-2026 school year, the Austin Independent School District documented roughly 20 separate instances of driverless Waymo robotaxis illegally passing stopped school buses that had their red lights flashing and stop arms extended.

02 // KEY FACTS

Case telemetry

INCIDENT
SS-IR-068
DATE
December 8, 2025
SYSTEM
Waymo
LOCATION / SCOPE
Austin, Texas and Atlanta, Georgia, United States
EVIDENCE
Reported
AI ROLE
Autonomous actor
HARM
Physical safety
SOURCES
3 cited records
03ENTRY POINT // WHAT HAPPENED

The event

During the 2025-2026 school year, the Austin Independent School District documented roughly 20 separate instances of driverless Waymo robotaxis illegally passing stopped school buses that had their red lights flashing and stop arms extended. Dashcam and district video showed the cars rolling past buses while children were disembarking; in one case a Waymo passed only moments after a student had crossed in front of it, while the student was still in the road. The pattern echoed an earlier Atlanta incident in which a Waymo drove around a stopped bus's stop arm and safety device with no human safety operator present. NHTSA opened a probe in October 2025 and expanded it in a letter dated December 3, 2025. In early-to-mid December 2025 Waymo filed a voluntary software recall covering 3,067 vehicles (production dates August 20 to November 5, 2025), which the company said had already been updated by November 17, 2025. The fix did not fully hold: at least five of the illegal passes occurred AFTER Waymo assured the district the problem had been resolved, and a 20th Austin citation was logged on December 1, 2025, weeks after the claimed software fix. No injuries were reported.

04CAUSAL TRACE // AI'S ACTUAL ROLE

What the machine did

Waymo's fully autonomous driving system was solely responsible for perceiving and obeying the stopped-school-bus rule, with no human safety operator in the vehicle and no per-trip human approval gate. The system repeatedly failed to recognize the legally controlling combination of flashing red lights and an extended stop arm and chose to navigate around the bus at machine speed. The deeper failure was in change validation: Waymo deployed a software update by November 17, declared the issue fixed to the school district, and let the cars keep driving past stopped buses. With no independent human SME confirmation that the new behavior actually held against real stopped-bus scenarios before re-attesting safety, at least five further illegal passes and a December 1 citation slipped through behind a fix that had not been proven to work.

Autonomous actorAutomation was a causal participant—not a decorative label for the system around it.
05BLAST RADIUS // CONSEQUENCES

Where the failure landed

NHTSA expanded its federal safety investigation into Waymo. Waymo issued a voluntary software recall affecting 3,067 robotaxis. Austin ISD recorded roughly 20 illegal passes of stopped school buses, each a documented child-safety hazard, including one pass while a student remained in the road. The reputational damage was compounded by the recall's incompleteness: the company had publicly claimed the issue was fixed, yet violations continued after the update, undermining trust in its self-attested safety improvements among regulators, school officials, and parents.

06 // EVIDENCE STATUS

Reported

Documented in the cited public record. Follow the sources for the precise evidentiary posture.

SOURCE RECORD UPDATED 2026-07-09

07 // SOURCE LEDGER

3 cited records

  1. 01
  2. 02
  3. 03
08CONTROL FAILURE // MISSING GOVERNANCE

Execution gate and human override

The failure pattern in this case: Autonomous high-consequence action.

09INTERVENTION POINT // HUMAN IN THE MIDDLE

The moment the path could change

A trained operator receives the evidence, owns the go/no-go decision, and retains an immediate override.

AI PROPOSES→HUMAN OWNS THE DECISION→SYSTEM EXECUTES
10CONTROL DEPLOYMENT // AUTHORITYGATE

High-consequence gate · human override

AuthorityGate's Operational Resilience framework requires a human SME change-validation gate before any safety-critical behavior change can be declared fixed and re-attested to a regulator or the public. A qualified transportation-safety SME would have to sign off on documented evidence that the updated stopped-school-bus logic was validated against the exact failing scenarios -- flashing reds plus extended stop arm, students in or near the roadway -- across a representative test fleet, with a defined pass threshold and a holdback period before the "resolved" status could be communicated to the school district or NHTSA. The November 17 update could not have been marked closed, and Waymo could not have told the district the problem was solved, until a human reviewer confirmed zero illegal passes in validation. The five post-fix incidents and the December 1 citation would have been caught at the validation gate and kept the matter open instead of being papered over by an unverified claim.

RELEVANT KEYSTONE CONTROLHuman-in-the-Loop ValidationHow high-risk actions route to a named subject-matter expert who owns the go or no-go decision.
12 // THE ALTERNATIVE

Autonomy is a design choice.

See the operating model that keeps AI useful while preserving human authority at consequential moments.

Compare AgenticAI and AugmentedAI →